Tardive dyskinesia remains a major public health problem in psychiatry. No proven safe and effective treatment has emerged for this condition and at present all marketed antipsychotic drugs appear capable of producing this adverse effect. Therefore, identification of risk factors and the development of strategies for prevention remain major goals. Knowledge regarding factors which might increase the risk for TD is largely based on cross- sectional prevalence surveys and retrospective data collection on important diagnostic, illness and treatment history characteristics. The continuation of the present prospective study would provide a unique opportunity to further develop our prospectively collected database on the incidence and course of tardive dyskinesia as well as factors influencing both of these. To date 955 patients have entered the study including 70 schizophrenic or schizophreniform patients with no prior exposure to neuroleptics at study entry and 135 patients receiving different controlled dosage ranges of fluphenazine decanoate. The median length of total previous neuroleptic exposure at study entry (for those who have been treated) is 12 months, therefore, patients who are being followed from the early part of their treatment course. Sixty-three patients with a documented history of severe parkinsonian side effects to neuroleptic medications are also being followed. The design of the study calls for patients to be evaluated every 3 months with the Simpson Dyskinesia Scale, the Simpson-Angus Scale and the BPRS. Patients with questionable signs of TD are seen monthly. Those developing presumptive TD are evaluated neuromedically, videotaped and seen every 2 weeks, with discontinuation of neuroleptic medication whenever possible. To date 160 cases of presumptive TD have developed and a major focus of the project at present is to investigate TD outcome and to evaluate the effect of prospectively assessed variables (both prior to and subsequent to the development of tardive dyskinesia) on its long-term course.